FHWA/TX-03/ Title and Subtitle ENGINEERING COUNTERMEASURES TO REDUCE RED-LIGHT- RUNNING. August Performing Organization Code

Transcription

1 1. Report No. FHWA/TX-03/ Title and Subtitle ENGINEERING COUNTERMEASURES TO REDUCE RED-LIGHT- RUNNING Technical Report Documentation Page 2. Government Accession No. 3. Recipient's Catalog No. 5. Report Date August Performing Organization Code 7. Author(s) James Bonneson, Karl Zimmerman, and Marcus Brewer 9. Performing Organization Name and Address Texas Transportation Institute The Texas A&M University System College Station, Texas Sponsoring Agency Name and Address Texas Department of Transportation Research and Technology Implementation Office P.O. Box 5080 Austin, Texas Performing Organization Report No. Report Work Unit No. (TRAIS) 11. Contract or Grant No. Project No Type of Report and Period Covered Research: September August Sponsoring Agency Code 15. Supplementary Notes Research performed in cooperation with the Texas Department of Transportation and the U.S. Department of Transportation, Federal Highway Administration. Research Project Title: Signalization Countermeasures to Reduce Red-Light-Running 16. Abstract Red-light-running is a significant problem throughout the United States and Texas. It is associated with frequent and severe crashes. Engineering countermeasures represent a useful means of combating the redlight-running problem because they are passively applied (in contrast to enforcement countermeasures which are considered to be overt and punitive) and are in the direct control of the agency responsible for the signal. The objective of this research project was to describe how engineering countermeasures can be used to minimize the frequency of red-light-running and associated crashes at intersections. This report documents the work performed, findings, and conclusions reached as a result of a two-year research project. During the first-year, engineering countermeasures were identified and implemented at 10 intersections in five Texas cities. Before-after studies of red-light-running frequency were then conducted at each intersection. Also, the three-year crash history for each intersection was compared to its observed frequency of red-light-running. The findings from these studies indicate that the frequency of red-lightrunning decreases in a predictable way with decreasing approach flow rate, longer clearance path lengths, longer headways, and longer yellow interval durations. The crash data analyses indicate that right-angle crashes increase exponentially with an increasing frequency of red-light-running. Models for computing an intersection approach s red-light-running frequency and related crash rate are described. Guidelines for selecting appropriate engineering countermeasures and evaluating their performance are provided. 17. Key Words Signalized Intersections, Change Interval, Yellow Interval, Red-Light-Running 19. Security Classif.(of this report) Unclassified Form DOT F (8-72) 20. Security Classif.(of this page) Unclassified Reproduction of completed page authorized 18. Distribution Statement No restrictions. This document is available to the public through NTIS: National Technical Information Service 5285 Port Royal Road Springfield, Virginia No. of Pages Price

2

3 ENGINEERING COUNTERMEASURES TO REDUCE RED-LIGHT-RUNNING by James Bonneson, P.E. Associate Research Engineer Texas Transportation Institute Karl Zimmerman Graduate Research Assistant Texas Transportation Institute and Marcus Brewer Associate Transportation Researcher Texas Transportation Institute Report Project Number Research Project Title: Signalization Countermeasures to Reduce Red-Light-Running Sponsored by the Texas Department of Transportation In Cooperation with the U.S. Department of Transportation Federal Highway Administration August 2002 TEXAS TRANSPORTATION INSTITUTE The Texas A&M University System College Station, Texas

4

5 DISCLAIMER The contents of this report reflect the views of the authors, who are responsible for the facts and the accuracy of the data published herein. The contents do not necessarily reflect the official view or policies of the Federal Highway Administration (FHWA) and/or the Texas Department of Transportation. This report does not constitute a standard, specification, or regulation. It is not intended for construction, bidding, or permit purposes. The engineer in charge of the project was James Bonneson, P.E. # NOTICE The United States Government and the State of Texas do not endorse products or manufacturers. Trade or manufacturers names appear herein solely because they are considered essential to the object of this report. v

6 ACKNOWLEDGMENTS This research project was sponsored by the Texas Department of Transportation (TxDOT) and the Federal Highway Administration. The research was conducted by Dr. James A. Bonneson, Mr. Karl Zimmerman, and Mr. Marcus Brewer with the Design and Operations Division of the Texas Transportation Institute. The researchers would like to acknowledge the support and guidance provided by the project director, Mr. Wade Odell, and the members of the Project Monitoring Committee, including: Mr. Baltazar Avila, Mr. Dale Barron, Mr. Mike Jedlicka, Mr. James Mercier, Mr. Doug Vanover, Mr. Roy Wright (all with TxDOT), and Mr. Walter Ragsdale (with the City of Richardson). In addition, the researchers would like to acknowledge the assistance provided by Mr. Ismael Soto (with TxDOT) in locating several field study sites and in implementing selected countermeasures. Finally, the valuable assistance provided by Dr. Montasir Abbas in the conduct of this research is also gratefully acknowledged. vi

7 TABLE OF CONTENTS Page LIST OF FIGURES....ix LIST OF TABLES.... x CHAPTER 1. INTRODUCTION OVERVIEW RESEARCH OBJECTIVE RESEARCH SCOPE RESEARCH APPROACH CHAPTER 2. RED-LIGHT-RUNNING PROCESS AND COUNTERMEASURES OVERVIEW EXPOSURE FACTORS CONTRIBUTORY FACTORS FACTORS LEADING TO CONFLICT RED-LIGHT-RUNNING COUNTERMEASURES CHAPTER 3. MODEL DEVELOPMENT AND COUNTERMEASURE SELECTION OVERVIEW MEASURES OF EFFECTIVENESS MODEL DEVELOPMENT COUNTERMEASURES TO BE EVALUATED CHAPTER 4. STUDY SITE SELECTION AND DATA COLLECTION OVERVIEW SITE SELECTION DATA COLLECTION PLAN CHAPTER 5. DATA ANALYSIS OVERVIEW ANALYSIS OF FIELD DATA ANALYSIS OF CRASH DATA CHAPTER 6. CONCLUSIONS OVERVIEW SUMMARY OF FINDINGS CONCLUSIONS vii

8 TABLE OF CONTENTS (Continued) Page CHAPTER 7. REFERENCES APPENDIX: GUIDELINES FOR SELECTING AND EVALUATING ENGINEERING COUNTERMEASURES TO REDUCE RED-LIGHT-RUNNING... A-1 viii

9 LIST OF FIGURES Figure Page 2-1 Effect of Flow Rate on the Frequency of Red-Light-Running Effect of Flow Rate and Detection Design on Max-Out Probability Probability of Stopping as a Function of Travel Time and Control Type Probability of Stopping as a Function of Travel Time and Speed Probability of Stopping as a Function of Travel Time and Approach Grade Probability of Stopping as a Function of Travel Time and Yellow Duration Probability of Stopping as a Function of Travel Time and Proximity of Other Vehicles Variation of Red-Light-Running and Other Conflicts by Time-of-Day Relationship between Probability of Stopping and Yellow Interval Duration Relationship between Red-Light-Running Frequency and Yellow Duration Probability of Going at Yellow Onset Frequency of Red-Light-Running as a Function of Time into Red Effect of an Increase in Yellow Interval Duration on the Frequency of Red-Light-Running Red-Light-Running Frequency as a Function of Approach Flow Rate Red-Light-Running Frequency as a Function of Flow-Rate-to-Cycle-Length Ratio Red-Light-Running Frequency as a Function of Yellow Interval Duration Red-Light-Running Frequency as a Function of Speed Red-Light-Running Frequency as a Function of Clearance Path Length Red-Light-Running Frequency as a Function of Platoon Ratio Red-Light-Running Frequency as a Function of Back Plate Use Prediction Ratio versus Predicted Red-Light-Running Frequency Comparison of Observed and Predicted Red-Light-Running Frequency Effect of a Change in Cycle Length on Red-Light-Running Effect of a Change in Yellow Interval Duration on Red-Light-Running Effect of a Change in Average Running Speed on Red-Light-Running Effect of a Change in Clearance Path Length on Red-Light-Running Effect of a Change in Platoon Ratio on Red-Light-Running Red-Light-Running Frequency as a Function of Yellow Interval Difference Predicted Effect of Yellow Duration and Speed on Red-Light-Running Frequency Comparison of Observed and Predicted Intersection Crashes Predicted Effect of Red-Light-Running on Intersection Crash Frequency Effect of a Change in Red-Light-Running on Crash Frequency ix

10 LIST OF TABLES Table Page 2-1 Events Leading to Red-Light-Running and Related Crashes Factors Affecting Driver Decision at Onset of Yellow Indication Relationship between Countermeasure Category and Driver Decision Type Engineering Countermeasures to Red-Light-Running Red-Light-Running-Related Measures of Effectiveness Intersection Characteristics Study Site Characteristics Countermeasure Implemented at Each Study Site Study Site Characteristics by Study Period Database Elements and Data Collection Method Database Summary - Total Observations Database Summary - Speed Statistics and Yellow Intervals Database Summary - Statistics for Selected Variables Red-Light-Running Rates at Each Study Site Calibrated Red-Light-Running Model Statistical Description Expected Red-Light-Running during the Before and After Periods Countermeasure Effectiveness Crash Frequency at Each Study Site Calibrated Crash Model Statistical Description Effect of Selected Variables on the Frequency of Red-Light-Running Engineering Countermeasures to Red-Light-Running x

11 CHAPTER 1. INTRODUCTION OVERVIEW Statistics indicate that red-light-running has become a significant safety problem throughout the United States. Retting et al. (1) report that about one million collisions occur at signalized intersections in the U.S. each year. Of these collisions, Mohamedshah et al. (2) estimate that at least 16 to 20 percent can be attributed directly to red-light-running. Retting et al. also report that motorists involved in red-light-running-related crashes are more likely to be injured than those in other crashes. In fact, they found that 45 percent of red-light-running-related crashes involve injury whereas only 30 percent of other crashes involve injury. A 1998 survey of Texas drivers, conducted by the Federal Highway Administration (FHWA) (3), found that two of three Texans witness red-light-running every day. About 89 percent of these drivers believe that red-light-running has worsened over the past few years. The largest percentage (66 percent) perceive the reason for red-light-running is that the red runner is in a hurry. An examination of nationwide fatal crash statistics by the Insurance Institute for Highway Safety found that Texas had the fourth highest number of red-light-running-related deaths per 100,000 population between 1992 and 1998 (4). There is a wide range of potential countermeasures to the red-light-running problem. These solutions are generally divided into two broad categories: engineering countermeasures and enforcement countermeasures. Enforcement countermeasures are intended to encourage drivers to adhere to the traffic laws through the threat of citation and possible fine. In contrast, engineering countermeasures (which include any modification, extension, or adjustment to an existing traffic control device) are intended to reduce the chances of a driver being in a position where he or she must decide whether or not to run the red indication. Studies by Retting et al. (1) have shown that countermeasures in both categories are effective in reducing the frequency of red-light-running. However, most of the research conducted to date has focused on the effectiveness of enforcement; little is known about the effectiveness of many engineering countermeasures. In summary, red-light-running is a significant problem throughout the United States and Texas. It appears to be a growing problem that leads to frequent and severe crashes. Engineering countermeasures represent an attractive means of combating the red-light-running problem because they are passively applied (in contrast to enforcement countermeasures which are considered to be overt and punitive) and are in the direct control of the agency responsible for the signal. This report describes the factors that are associated with red-light-running as well as several countermeasures that have been used to reduce its frequency. Initially, there is an examination of the red-light-running process in terms of the events necessary to precipitate a red-light-running event. Then, various engineering countermeasures are identified. Next, a before-after study is described. This study is intended to facilitate the evaluation of selected countermeasures and to calibrate a 1-1

12 model for predicting the frequency of red-light-running. The data are then analyzed and the findings used to develop guidelines for selecting and evaluating engineering countermeasures. RESEARCH OBJECTIVE The objective of this research project was to describe how engineering countermeasures could be used to minimize the frequency of red-light-running and associated crashes at intersections. Satisfying of the following goals helped achieve this objective:! Quantify the effect of various traffic characteristics and traffic control factors on the frequency of red-light-running.! Quantify the relationship between red-light-running and crash frequency.! Identify promising engineering countermeasures and quantify their effects.! Facilitate implementation of engineering countermeasures through development of a guide. RESEARCH SCOPE A red-light-running event can be characterized by traffic movement type, entry time of the red-light-running vehicle, and the motivation underlying the driver s decision to run the red indication. Traffic movement type reflects the different expectations and experiences of the leftturn versus the through driver. Relative to the through driver, the left-turn driver is forced (by geometry) to travel through the intersection at a slow rate of speed and is also more likely to experience lengthy delays. Entry time of the red-light-running driver relates to the time that the driver enters the intersection after the onset of the red indication. When a driver enters late into the red, it may be an indication of deficiencies in signal visibility or driver sight-distance along the intersection approach. It may also be an indicator of driver indifference to the traffic laws regarding the red indication. Intuitively, crash potential is higher when a red-light-runner enters several seconds after the red onset. Fortunately, about 85 percent of all red-light-runners enter the intersection within the first 1.5 s of red (5) so the frequency of related crashes due to late entries is relatively low. Driver Decision Type describes the basis for the driver s decision to run the red indication. An avoidable red-running event is committed by a driver who believes that it is possible to safely stop but decides it is in his or her best interest to run the red indication. In contrast, an unavoidable event is committed by a driver who either (1) believes that he or she is unable to safely stop and consciously decides to run the red, or (2) is unaware of the need to stop. This research focuses on the unavoidable red-light-running by through drivers that takes place during the first few seconds after the onset of red. Red-light-running events having these characteristics occur frequently and are most treatable by engineering countermeasures. Moreover, efforts to reduce this type of red-light-running are likely to have the greatest return in terms of a reduced number of crashes. 1-2

13 RESEARCH APPROACH The research approach is based on a two-year program of development and evaluation that was directed at producing information engineers could use to reduce red-light-running. During the first year of the research, the project team identified causes of red-light-running and a range of engineering countermeasures. Researchers developed a before-after study plan to evaluate the effectiveness of alternative countermeasures at 10 signalized intersections in Texas. In the second year, several countermeasures were implemented and evaluated through the direct measurement of red-light-running frequency. The project compared the crash history of the study sites to the observed frequency of red-light-running. One product of this research is a guideline document. This document provides technical guidance for engineers interested in using engineering countermeasures to reduce red-light-running at problem intersections. It also provides tools for evaluating the effectiveness of selected engineering countermeasures. 1-3

14

15 CHAPTER 2. RED-LIGHT-RUNNING PROCESS AND COUNTERMEASURES OVERVIEW This chapter describes the red-light-running process and the countermeasures described in the literature as having some effect on the frequency of red-light-running. Initially, the red-lightrunning process is described in terms of the events that lead to red-light-running and the factors that have some influence on a driver s propensity to run the red indication. The chapter concludes with a discussion of red-light-running countermeasures, with a focus on engineering countermeasures. Red-Light-Running Process Several events must occur together to result in a driver running the red indication. Additional events must then follow for a red-light-running-related crash to occur. Table 2-1 lists these events in roughly the same sequence that they must occur to produce a red-run event and subsequent crash. Table 2-1. Events Leading to Red-Light-Running and Related Crashes. Type Event RLR Freq. Exposure Events Contributory Events Rt. Angle Crash Rear-end Crash 1. Vehicle iis xsec. travel time from the intersection (x <6.0 s). U U U 2. Phase terminates (yellow presentation). U U U 3. Phase termination is by phase max-out (or controller is pretimed). U U U 4. Vehicle idoes not stop. U U U 5. Vehicle i s entry time occurs after yellow ends. U U 6. Vehicle i s clearance time occurs after all-red ends. U 7. Conflicting vehicle kenters intersection ysec. after all-red ends. U 8. Vehicle jstops (and it is in front of vehicle i). U Note: RLR =red-light-running The first three events listed in Table 2-1 represent exposure events because they set the stage for the contributory events that follow. Thus, exposure to red-light-running requires: (1) sufficient traffic volume to result in one or more vehicles on the intersection approach; (2) a phase termination; and (3) pretimed control or, if the control is actuated and advance detection is used, the termination is by max-out (i.e., maximum green limit is reached). Consideration of the first two events suggests that exposure to red-light-running increases with flow rate on the subject approach and the number of signal cycles. 2-1

16 The contributory events that lead to red-light-running include: (1) the vehicle does not stop, and (2) the vehicle s time of entry into the intersection occurs after the indication changes from yellow to red. Consideration of these two events suggests that the frequency of red-light-running will increase whenever drivers are less likely to stop and when the yellow interval is reduced. The vehicle does not stop event is the most complex event of those listed in Table 2-1. The probability of this event is discussed herein in terms of its inverse, the probability of stopping. It reflects the uncertainty (or indecision) exhibited by the population of drivers on an intersection approach at the onset of the yellow indication. The event is complex because many factors can affect the probability of stopping (e.g., travel time to intersection, speed, etc.). The last two columns of Table 2-1 relate to the two types of crashes most commonly found at signalized intersections. Both types require the same exposure events. The right-angle crash also requires: (1) the red-light-running vehicle to be present in the intersection when the all-red interval ends, and (2) a conflicting vehicle to enter the intersection while it is occupied by the red-lightrunning vehicle. Consideration of these two events suggests that the frequency of right-angle crashes increases with a decrease in the all-red interval and an increase in the conflicting movement flow rate. In summary, the following factors influence the frequency of red-light-running and related crash frequency:! flow rate on the subject approach (exposure factor),! number of signal cycles (exposure factor),! phase termination by max-out (exposure factor),! probability of stopping (contributory factor),! yellow interval duration (contributory factor),! all-red interval duration (contributory factor),! entry time of the conflicting driver (contributory factor), and! flow rate on the conflicting approach (exposure factor). Each of these factors is described more fully in a later section of this chapter. Review of Texas Law To provide some perspective on the problem of red-light-running in Texas, it is important to be familiar with the applicable laws, codes, and ordinances. Chapter 544 of the Texas Transportation Code (6) deals with traffic signs, signals, and markings; section specifically addresses traffic-control signals. Section is somewhat ambiguous concerning the specific problem of red-lightrunning, in that the definition of a driver s responsibility when encountering a yellow signal is not fully specified. According to Subsection (b), a driver waiting at an intersection when his or her 2-2

17 signal turns green must wait until all other legally entering vehicles have cleared the intersection before proceeding. Therefore, Texas law implies that a vehicle that enters an intersection legally (i.e., during yellow) may still be in the intersection after a conflicting movement receives a green signal. This law is sometimes referred to as the permissive yellow law in comparison to more restrictive laws that require drivers to have exited the intersection before the end of the yellow interval. Parsonson et al. (6) indicate that at least half of the states in the United States follow the permissive rule. The advantage of the permissive rule is that it enables most drivers to be lawful in their responses to the yellow indication. The disadvantage of the permissive law is that it creates a situation where the cross-street driver receives a green indication but must yield the right-of-way (to a crossing vehicle) before entering the intersection. Parsonson et al. (6) indicate that 60 percent of drivers are unaware that they have to yield the right-of-way when presented the green indication. Moreover, when asked the question, What would you think if traffic engineers decided to time yellow lights so that there might be a vehicle going through the intersection when you get your green, 69 percent of drivers said that they disapproved of this practice because it was dangerous. The solution advocated by Parsonson et al. (6) is to provide an all-red interval (following the yellow interval) of sufficient duration to permit drivers to clear the intersection before a conflicting phase is presented with a green indication. EXPOSURE FACTORS This section summarizes the literature as it relates to events that expose drivers to conditions that may precipitate red-light-running. These events were previously discussed with regard to Table 2-1. The factors that underlie these events include flow rate, number of signal cycles, and phase termination by max-out. Flow Rate on the Subject Approach Flow rate on the subject approach is important to the discussion of red-light-running. Each vehicle on the intersection approach at the onset of yellow is exposed to the potential for red-lightrunning. The number of drivers running the red each signal cycle will likely increase as the flow rate increases. Three studies have reported sufficient data to examine the effect of flow rate on red-lightrunning frequency or related crashes. Kamyab et. al (7) observed 1242 hours of operation at 12 urban intersections in Iowa. For each intersection approach, they reported the average daily traffic volume, the observed number of red-light-runners, and the duration of the study. The relationship between the computed hourly approach flow rate and red-light-running frequency is shown in Figure 2-1 (with square data points). This trend line indicates that red-light-running increases at a rate of about 3.0 red-light-runners per 1000 vehicles. 2-3

18 Red-Light-Running Frequency, veh/h Baguley (8 ) outlier Kamyab et al. (7 ) Approach Flow Rate, veh/h Figure 2-1. Effect of Flow Rate on the Frequency of Red-Light-Running. Baguley (8) examined the frequency of red-light-running at seven rural intersections in England. He found that red-light-running frequency was positively correlated with approach flow rate. He also noted that there was a slight positive correlation with approach speed and inverse correlation with daily cross-street volume. The relationship between flow rate and red-light-running frequency using Baguley s (8) data is shown in Figure 2-1 (using circular data points). Six intersections (shown with solid circles) had daily cross-street volumes of less than 7500 vehicles. The seventh intersection (shown with an open circle and labeled outlier ) had an exceptionally high daily cross-street volume of 17,000 vehicles that Baguley speculated may explain its very low frequency of red-light-running. The trend line indicates that red-light-running increases at a rate of about 5.3 red-light-runners per 1000 vehicles. Mohamedshah et al. (2) examined the effect of flow rate (and other variables) on red-lightrunning-related crashes. They obtained crash data for 1756 urban intersections in California. The data were screened to include only those crashes attributable to a red-light-running event. They found that crash frequency increased with flow rate on the subject approach. Their findings indicate that approach crash frequency increases from 0.25 crash/yr at a two-way volume of 8000 veh/day to 0.5 crash/yr at 50,000 veh/day. Number of Signal Cycles As noted previously, some researchers recognize that the frequency of red-light-running and related crashes is largely affected by the frequency with which the yellow indication is presented (9). 2-4

19 A cycle length change from 60 to 120 s reduces the number of times that the yellow is presented by 50 percent. In theory, a similar reduction in red-light-running frequency should also be observed. Recognition of this relationship is often exhibited by the researchers reporting red-light-running statistics normalized by cycle frequency. For example, Van der Horst and Wilmink (9) discuss the use of percent of cycles with at least one red-light-runner. Phase Termination by Max-Out Green-extension detection systems use one or more detectors located in advance of the intersection to hold the phase in green as long as the approach is occupied. In this manner, drivers are not exposed to the yellow indication, and red-light-running is reduced. However, if the green is held to its maximum limit, the phase maxes-out and is forced to end, regardless of whether a vehicle is approaching the intersection. An actuated phase that maxes-out has the potential to expose drivers to a red-light-running situation. Similarly, a pretimed signal phase always ends independently of vehicle presence on the approach and has the potential to expose drivers to a red-light-running situation. Zegeer and Deen (10) evaluated the effect of green-extension systems on the frequency of red-light-running. Their evaluation focused on two rural intersections. It revealed a 65 percent reduction in red-running frequency due to the use of a green-extension system. As noted previously, the benefits of a green-extension system can be negated if the phase maxes-out. The probability of max-out is dependent on flow rate in the subject phase and the maximum allowable headway, as dictated by the detector design. The maximum allowable headway (MAH) is the largest headway in the traffic stream that can occur and still sustain a continuous extension of the green interval. The relationship between max-out probability, MAH, maximum green, and flow rate is illustrated in Figure 2-2. Bonneson and McCoy (11) indicate that the MAH values shown in Figure 2-2 (i.e., 4.0 and 7.0 s) represent the range of values for most detection designs. To illustrate the implications of alternative detection designs, consider the following example. If a phase has a flow rate of 1200 veh/hr, a maximum green duration of 30 s, and no advance detection (i.e., only a stop-line detector), then its probability of max-out will be about 0.05 (1 out of 20 cycles). However, if a green-extension system is used, the resulting max-out probability will increase to 0.7 (7 out of 10 cycles). One option available to reduce this probability is to increase the maximum green setting; however, this increase may also increase the delay to waiting vehicles. 2-5

20 Probability of Max-Out s Maximum Green 50-s Maximum Green Advance Detection (7.0 s MAH) No Advance Detection (4.0 s MAH) Total Flow Rate in Subject Phase, veh/h Figure 2-2. Effect of Flow Rate and Detection Design on Max-Out Probability. CONTRIBUTORY FACTORS Two contributory factors underlie the events that lead to red-light-running. These factors include the probability of stopping and the yellow interval duration. The former factor represents the complex decision-making process that drivers exhibit at the onset of the yellow indication. A review of the literature indicates that this decision is affected by the driver s assessment of the prevailing traffic and roadway conditions. It is also affected by the driver s estimate of the consequences of stopping (or not stopping). The yellow interval duration contributes to red-light-running in a more fundamental manner. The start of this interval defines the instant when the decision-making process should begin. During this interval, a decision is made and acted upon. The end of this interval defines the instant when the red indication is presented (whereupon entry to the intersection represents a red-light-running event). Both factors, and their relationship to the frequency of red-light-running, are described in this section. Probability of Stopping Many researchers have studied the decision to stop in response to the yellow indication. Van der Horst and Wilmink (9) studied this decision process and found that a driver s propensity to stop is based on three components. These components and the factors that influence them are listed in Table 2-2. Each component is discussed in the following subsections. 2-6

21 Table 2-2. Factors Affecting Driver Decision at Onset of Yellow Indication. Components of the Decision Process Factor Driver behavior Travel time Speed Actuated control Headway Coordination Approach grade Yellow interval Estimated consequences of not stopping Estimated consequences of stopping Threat of right-angle crash Threat of citation Threat of rear-end crash Expected delay Driver Behavior In the case of an unavoidable red-light-running event, the driver s response to the yellow indication is affected by his or her perceived ability to stop and his or her awareness of the need to stop. An unavoidable event is committed by a driver who either (1) believes that he or she is unable to safely stop and consciously decides to run the red indication, or (2) is unaware of the need to stop. This ability and awareness is influenced by the seven factors listed in Table 2-2. Each of these factors is discussed in the following paragraphs. Travel Time. Studies indicate that a driver s decision to stop at yellow onset is based partly on his or her estimate of speed and distance to the stop line (12, 13, 14, 15, 16, 17). Through these estimates, the driver assesses his or her ability to stop and the degree of comfort associated with the stop. Several researchers have measured driver response to the yellow indication in terms of the travel time to the intersection at the onset of yellow (9, 12, 13, 14, 15, 16). The relationship between travel time and probability of stopping reported by each researcher is shown in Figure Olson & Rothery (12) Probability of Stopping Williams (13) Hulsher (14) Sheffi & Mahmassani (15) 30 mph Actuated Pretimed Bonneson et al. (16) Van der Horst & Wilmink (9 ) Travel Time to Stop Line, s Figure 2-3. Probability of Stopping as a Function of Travel Time and Control Type. 2-7

22 The trends in Figure 2-3 indicate that there is a range, between about 2- and 5-s travel time from the intersection stop line, where drivers are collectively indecisive about the decision to stop. The solid and dashed lines suggest that there is a difference in driver behavior at pretimed and at actuated intersections. This trend is discussed in a subsequent section titled Actuated Control and Coordination. Speed. A driver s decision to stop may be skewed by his or her limited ability to estimate travel time to the intersection at higher speeds. Allsop et al. (17) found that drivers tend to underestimate actual travel time by about 30 percent. Related to this observation is the reported finding that high-speed drivers tend to be less likely to stop than low-speed drivers when at the same travel time from the stop line at the onset of the yellow indication (15, 16). The trend reported by Bonneson et al. (16) is shown in Figure 2-4. Figure 2-4. Probability of Stopping as a Function of Travel Time and Speed. The trends shown in Figure 2-4 suggest that the time interval within which drivers are indecisive varies slightly with approach speed. Drivers that are 4.0 s from the stop line have a 0.6 probability of stopping if they are traveling at 35 mph; however, they have only a 0.2 probability if they are traveling at 55 mph. This behavior suggests that the degree to which a driver underestimates his or her travel time increases with speed. 1.0 Probability of Stopping mph 55 mph Travel Time to Stop Line, s Actuated Control. Evidence of the effect of intersection control type on the probability of stopping has been reported by Van der Horst and Wilmink (9). They found evidence that drivers approaching an actuated intersection are less likely to stop than if they are approaching a pretimed intersection. This finding suggests that drivers learn which signals are actuated and then develop 2-8

23 Coordination. Van der Horst and Wilmink (9) extrapolated the aforementioned driver expectancy associated with actuated control to drivers traveling within platoons through a series of interconnected signals. Drivers in a platoon are believed to develop an ad hoc expectancy as they travel without interruption through successive signals. Their expectancy is that each signal they approach will remain green until after they (and the rest of the platoon) pass through the intersection. Their desire to stay within the platoon makes them less willing to stop at the onset of the yellow indication. Approach Grade. Chang et al. (18) examined the effect of approach grade on the probability of stopping. They found that drivers on downgrades were less likely to stop (at a given travel time from the stop line) than drivers on level or upgrade approaches. The effect of grade is shown in Figure 2-5 for an approach speed of 30 mph. The trends in this figure suggest that only about 38 percent of drivers will stop on a 5 percent downgrade when they are 4 s travel time from the stop line. In contrast, 66 percent will stop if they are on a 5 percent upgrade. an expectation of service as they travel through the detection zone. This effect of control type on the probability-of-stopping is shown in Figure Probability of Stopping mph Grade = +5% (uphill) Travel Time to Stop Line, s Figure 2-5. Probability of Stopping as a Function of Travel Time and Approach Grade. Yellow Interval Duration. Van der Horst and Wilmink (9) have noted that long yellow intervals can lead to bad behavior because the last-to-stop drivers are not rewarded with a red indication as they arrive at the stop line. Instead, the yellow remains lit as they roll up to the stop line. These drivers will be more inclined not to stop the next time they approach the intersection. Several researchers have found that a driver adjusts his or her stopping behavior to offset the effect 2-9

24 of longer change intervals (8, 9, 19). This behavior is illustrated in Figure 2-6 and is based on the data reported by Van der Horst and Wilmink (9). This figure indicates that drivers that are 4.0 s from the stop line have a probability of 0.5 of stopping if the yellow is 3 s in duration; however, they have only a 0.34 probability if the yellow is 5 s long. 1.0 Actuated Control Probability of Stopping Yellow = 3 s Yellow = 4 s Yellow = 5 s Travel Time to Stop Line, s Figure 2-6. Probability of Stopping as a Function of Travel Time and Yellow Duration. Finally, a study by Mahalel and Prashker (19) indicates that a lengthy end-of-phase warning interval can lead to an increased indecision zone. Specifically, they found that when a 3-s yellow was preceded by a 3-s flashing green, the indecision zone ranged from 2 to 8 s. This range is larger than that for signals without a flashing green interval (i.e., 2 to 5 s). They cite evidence that an increased indecision zone increases the frequency of rear-end crashes. Headway. Drivers traveling through an intersection may be more cognizant of vehicles immediately ahead of them or just behind them than they are of the signal indication. Thus, they are likely to be drawn through the intersection by a preceding driver, even though the yellow (or red) indication is presented. In fact, Allsop et al. (17) found that drivers that are closely following (i.e., 2 s or less headway to the vehicle ahead) are more likely to run the red indication than are drivers that are neither closely following nor being closely followed (i.e., freely flowing drivers). An analysis of the data reported by Allsop et al. (17) is shown in Figure 2-7. The trends in this figure indicate that about 50 percent of drivers (at 3 s travel time from the stop line) are likely to stop if flowing freely on the approach. However, only about 42 percent of drivers will stop if they are within 2 s of the vehicle ahead. If these drivers are being closely followed, this percentage drops even further. This latter behavior is discussed in the section titled Consequences of Stopping. 2-10

25 1.0 Probability of Stopping Subject vehicle Free flow vehicle Travel Time to Stop Line, s Figure 2-7. Probability of Stopping as a Function of Travel Time and Proximity of Other Vehicles. Consequences of Not Stopping The driver s response to the yellow indication is also affected by his or her consideration of the consequences of not stopping and the consequences of stopping. The former consideration includes an estimate of the potential for a right-angle crash and the potential for receiving a citation. The latter consideration is discussed in a subsequent section titled Consequences of Stopping. Threat of Right-Angle Crash. A driver contemplating running the red indication may assess the threat of a right-angle crash by estimating the number of vehicles in the conflicting traffic stream. This number is estimated by scanning the intersection ahead and by recalling prior experience at this intersection. In this regard, a study by Baguley (8) found an inverse correlation between the frequency of red-light-running and the daily cross-street volume. His data indicate that drivers are six times more likely to run the red indication when the cross street has a daily traffic volume of 2000 veh/day compared to when it has 17,000 veh/day. Threat of Citation. Van der Horst and Wilmink (9) noted that drivers consider the potential for being cited when deciding whether to run the red indication. The findings from a survey of drivers, conducted by Retting and Williams (20), support this claim. They found that 46 percent of drivers (in cities without automated enforcement) believe that someone who runs the red indication is likely to receive a citation. This percentage increases to 61 percent in cities with automated enforcement. 2-11

26 Figure 2-7 shows the effect close following on the probability-of-stopping. The trends in this figure indicate that about 50 percent of drivers (at 3 s travel time from the stop line) are likely to stop if flowing freely on the approach. However, only about 25 percent of drivers will stop if they are being closely followed. This percentage drops to 8 percent when the driver is both closely followed and closely following another vehicle. Expected Delay. A survey conducted by the FHWA (3) indicated that 66 percent of Texas drivers believe red-light-running is due to drivers who are in a hurry. Obviously, the delay associated with stopping is contrary to most driver s desire to reach his or her destination quickly. A review of the literature did not uncover any research conducted on the effect of the drivers expected delay on the decision to stop at the onset of yellow. However, some evidence of this influence can be found in an examination of the data reported by Zegeer and Deen (10). These data include conflicts and flow rates observed throughout the day at two intersections, both before and after installation of a green-extension system (via multiple advance detectors). About two-thirds of the conflicts observed were red-light-runs. The relationship between conflict rate (in units of conflicts per 1000 vehicles ) and time-of-day is shown in Figure 2-8. The trends in Figure 2-8 indicate that drivers traveling during the noon and evening peak traffic hours are more likely to run a red light than during other hours of the day. This trend was exhibited in both the before and after periods. As delays tend to be highest during the peak hours, the trends suggest that drivers may be more inclined to run the red indication as the expected delay increases. Consequences of Stopping A driver s concern about the threat of a rear-end crash or lengthy delay is also factored into the decision to stop when presented with a yellow indication. Threat of Rear-End Crash. Drivers that are being closely followed when the light turns from green to yellow may be more reluctant to stop because of the greater likelihood of a rear-end crash. In a laboratory setting, Allsop et al. (17) observed that drivers being closely followed (i.e., when the following vehicle s headway was less than 2 s) at the onset of yellow were more likely to run the red indication. Yellow Interval Duration The yellow interval duration is generally recognized as a key factor that affects the frequency of red-light-running. This recognition has led several researchers to recommend setting the yellow interval duration based on the probability of stopping (9, 12, 18). These researchers suggest that the yellow interval should be based on the 85 th (or 90 th ) percentile driver s travel time to the stop line. This approach is illustrated in Figure 2-9 where the trends shown suggest that a yellow interval of 4.2 s is sufficient for 85 percent of drivers. Only 15 percent of drivers would choose to run the red 2-12

27 indication if they are more than 4.2-s travel time from the stop line at the onset of yellow and are in the first-to-stop-position. C onflict R ate, conflicts/1000 veh Before advance detection After advance detection Time of Day Figure 2-8. Variation of Red-Light-Running and Other Conflicts by Time-of-Day. 1.0 Run red with some risk. Probability of S topping Run red with significant risk. Yellow AR Next Green Travel Time to Stop Line, s Figure 2-9. Relationship between Probability of Stopping and Yellow Interval Duration. 2-13

28 A benefit of the all-red clearance interval is to provide a degree of protection against a rightangle conflict should a vehicle run the red indication. This benefit is realized if the all-red interval equals or exceeds the time required by the clearing vehicle to cross the intersection. Figure 2-9 illustrates the benefit of an all-red (AR) interval in terms of its ability to protect about two-thirds of the red-light-running vehicles from conflict (i.e., 10 of the 15 percent of all drivers that run the red). The trends in this figure suggest that if a 0.8-s all-red interval is used, then only 5 percent of drivers would be at significant risk for a right-angle conflict. FACTORS LEADING TO CONFLICT Once the driver has been presented the yellow indication and has chosen not to stop, there is a threat of conflict with other vehicles. This conflict can lead to a crash if one or both drivers are unable to effect an evasive maneuver. The frequency of a rear-end conflict (that occurs when the lead driver decides to stop and the following driver decides not to stop) is dependent on: (1) the probability of a red-light-running event, (2) the probability that two vehicles are present on the subject approach, and (3) the probability that the driver of the lead vehicle chooses to stop. The second probability is based on the flow rate on the subject approach as a contributing factor. The first and third probabilities were the subject of the preceding section. The frequency of a right-angle conflict is dependent on: (1) the probability of a red-lightrunning event, (2) the probability that a vehicle is present on the conflicting approach, and (3) the probability that it enters the intersection before the red-light-running vehicle clears it. The second and third probabilities are based on two contributing factors and one exposure factor. The contributing factors include the duration of the all-red clearance interval and the entry time of the conflicting driver. The exposure factor is the flow rate on the conflicting approach. These factors are discussed further in this section. All-Red Interval Duration The Manual on Uniform Traffic Control Devices (21) states that the yellow change interval may be followed by an all-red clearance interval to provide additional time before conflicting traffic movements are released. However, according to Parsonson et al. (6), there is no consensus at this time on whether this means that the clearance interval should be sufficiently long to completely clear the intersection or the degree to which the concept should be applied systemwide. This lack of a guidance has led to an inconsistency in the use of the all-red interval among agencies and may contribute to an increase in crashes due to driver confusion or a lack of driver respect for the signal. Entry Time of the Conflicting Driver The lead driver in a conflicting traffic stream could be in one of four states after receiving the green indication. These states are: (1) the driver is stopped at the stop line and pauses to verify that the intersection is clear before proceeding; (2) the driver is stopped at the stop line and tries to anticipate the onset of green by rolling forward during the all-red interval; (3) the driver is 2-14

29 approaching the intersection but is slowing to stop for the red interval; or (4) the driver is approaching the intersection but is anticipating the onset of green and maintains a nominal speed. The risk of conflict increases from State 1 to State 4. Any of the four states can occur; however, States 1 and 2 are most likely to occur at intersections when the flow rates are moderate to high. Researchers (14, 18) have examined the times associated with States 1 and 2 and found that almost all stopped, first-in-queue drivers require more than 1.0 s to reach the path of the clearing vehicle. This finding suggests that the red indication would have to be run and the clearing vehicle would have to be in the intersection 1.0 s or more after the conflicting movement receives the green for a conflict to occur. Hence, when flow rates are moderate to high, conflicting streams are separated in time even when an all-red interval is not provided. However, this protection may not extend to low-volume evening hours when States 3 or 4 are likely to occur. Flow Rate on the Conflicting Approach By definition, a conflict requires two or more vehicles to interact where one or more of these vehicles have to take an evasive action to avoid a collision. Thus, the frequency of conflict is a function of the flow rate of the conflicting traffic movements. As evidence of this effect, Mohamedshah et al. (2), in a study of red-light-running crash frequency, found that right-angle crashes on the major street increased with an increase in the flow rate on the minor street. RED-LIGHT-RUNNING COUNTERMEASURES There is a wide range of potential countermeasures to the red-light-running problem. These solutions are generally divided into two broad categories: engineering countermeasures and enforcement countermeasures. Enforcement countermeasures are intended to encourage drivers to adhere to the traffic laws through the threat of citation and possible fine. In contrast, engineering countermeasures are intended to reduce the frequency that drivers are put in a position where they must decide whether or not to run the red indication. The relationship between countermeasure category and the type of decision made by the driver (when running the red) is shown in Table 2-3. Table 2-3. Relationship between Countermeasure Category and Driver Decision Type. Driver Decision Type Possible Scenario Countermeasure Category Engineering Enforcement Avoidable Congested, Cycle overflow Less Most Effective Unavoidable Incapable of stop, Inattentive Most Effective Less The information in Table 2-3 suggests that there are two basic types of decisions associated with a red-light-running event. An avoidable red-running event is committed by a driver who believes that it is possible to safely stop but decides it is in his or her best interest to run the red 2-15